Research projects

Discover 7 Mac Eng research projects, from social bots to algorithmic watchdogs

The federal government invests in McMaster Engineering researchers, who work on exciting new developments in materials science, civil engineering, computer science and software.

McMaster researchers receive close to $11 million in funding for 51 research projects under the 2022 Discovery Research Program of the Natural Sciences and Engineering Research Council of Canada (NSERC).

Fourteen McMaster researchers, half of whom were McMaster engineering professionals, will also receive the Discovery Launch Supplement, which is awarded to early career researchers in the first year of their grant to help jump-start their careers. .

From social robots to improving biomedical materials, learn about the innovative work of seven McMaster Engineering researchers:

Sonia Hassini

Public works

Approach to sustainable stormwater management: optimal implementation of low-impact development systems

While the most widely used management practices for stormwater runoff have been end-of-pipe infrastructure, which diverts runoff to the nearest receiving water body, Canadian cities have recently adopted low impact. These technologies, such as rain gardens and green roofs, aim to manage runoff close to its source.

Because reliable observed data are scarce, there are levels of uncertainty about their functionality, performance, maintenance, placement, and collective long-term impact on large-scale stormwater management in the watershed.

This research will develop a framework that can help implement sustainable stormwater management in Canadian urban areas in the face of climate change and rapid urban development. The knowledge generated will also ensure the resilience and sustainability of Canadian communities in the face of extreme precipitation, climate change and urban growth for future generations.

Shahab Asoodeh

IT and software

Algorithmic Watchdog for Differential Privacy: From Theory to Practice

Powerful advances in machine learning can allow adversaries to gather personal information from an individual’s growing fingerprint. Despite the widespread adoption of Differential Privacy (DP), there is no standardized approach to evaluating and monitoring these technologies.

This research will create mathematical methods that accurately characterize the risks of private information leakage and reduced utility in existing DP algorithms. It will create a rigorous plan for scalable “algorithmic watchdogs” that monitor DP technologies for abuse and unintended damage.

In the long term, the research aims to help government and corporate data scientists with the tools needed to ensure meaningful and operational privacy.

Lingyang-chu

IT and software

Interpretations and actions towards reliable graph neural networks

Graph neural networks (GNNs) have been adopted by major companies – such as Google, Amazon, Facebook and Microsoft – to develop and achieve peak performance in enterprise applications. But most models are constructed and used as black boxes, which are not easily interpreted by humans.

This issue makes it difficult to verify and control GNN behaviors, poses serious security issues, reduces user trust, and prohibits mass deployment.

The research will develop reliable economic intelligence based on reliable and consistent interpretations of GNNs. It will advance the frontier in the growing field of trusted big data analytics.

Denise Geiskkovitch

IT and software

Designing robots for young children

As social robots emerge in environments with young children, recent research points out that interactions can lead to negative and inappropriate interactions, such as bullying and overconfidence.

While robots are designed and marketed for young children, says Geiskkovitch, the best way to do this is not known.

Using new methods to assess child-robot interaction with preschoolers, this research will advance the field by developing prescriptive frameworks to guide robot design and interaction behaviors to achieve results. positive.

This will be the first research program in Canada (and one of the few in the world) to study this area. The research will include user-centered workshops, laboratory experiments, and field studies to better understand young children’s expectations of social components of robots (e.g., usability), as well as logistical requirements (e.g. example, microphone placement) and outcomes (e.g., confidence). child-robot interaction.

Claudio Mengi

IT and software

Automated assistance for the design of cyberphysical systems: from theory to practice

Most of our industries, including automotive, energy, and healthcare, rely on cyber-physical systems for their day-to-day operations. But CPS design is complex, error-prone and therefore expensive, and failures can be catastrophic.

Engineers need automated support for CPS design, and despite many successes, there are still many challenges that prevent extensive use of these techniques in practice.

The research program will help engineers develop secure CPS by defining new software engineering solutions while targeting four challenges, including the need for:

  • Automated techniques that search for and detect flaws in CPS design
  • Helps to understand the causes of problems
  • Procedures that automatically translate human artifacts into machine-processable inputs
  • Comprehensive tools that help engineers to run different analysis on CPS design.

Yingying Wang

IT and software

Modeling diverse, custom, and expressive animations for virtual characters using motion capture, synthesis, and perception

Character animation plays a key role in game development, robotics, virtual reality, and augmented reality applications. Previous research into modeling movement to be natural and realistic has focused on generic movement content, meaning a lack of individual styles between characters.

This research will find solutions to generate stylized movements with variations that match real-world diversity to promote diversity, equity, and inclusion in the virtual world.

A major challenge for stylized movements is the lack of large scale movement style databases. This research will capture, learn and model three stylistic characteristics:

  • Demographic styles belonging to different groups of people, for example age, gender and race
  • Custom styles resulting from personalities
  • Bodybuilding
  • Expressive styles for various emotional and physical states for the same individual in different scenarios.

Kyla Saskatchewan

Materials Science and Engineering

Technical surfaces for the control of protein and cellular interactions and improved biomedical materials

When biomaterials encounter biological fluids such as blood, protein adsorption occurs rapidly and influences subsequent interactions with cells, including platelets, white blood cells, and microbes.

For medical devices such as catheters, stents, etc., these responses can lead to blood clotting, inflammation or infection, and ultimately device failure. For the devices to work, anticoagulant and antiplatelet drugs are always needed, and their use can lead to complications.

In order to improve biomaterials, surface modifications can be applied to alter their properties; physical modifications to control nanoscale properties and biofunctionalization may in turn allow control of cellular response and reduce adverse reactions.

This research will be directed towards a better understanding of protein and cell interactions with multifunctional materials using advanced surface modification methods for improved devices.

Thirteen other McMaster researchers received funding under the Discovery Stream:

  • Samir Chidiac
  • Douglas Down
  • Thomas Doyle
  • Shiva Kumar
  • Vladimir Mahalec
  • Prashant Mhaskar
  • Mehdi Moradi
  • Mehdi Narimani
  • Andre Phillion
  • Ravi Selvaganapathy
  • Sesha Srinivasan
  • Li Xi
  • Guxi